Probabilistic security-constrained AC optimal power flow

We propose a probabilistic framework for designing an N-1 secure dispatch for systems with fluctuating power sources. This could be used in various optimal power flow related applications, however in this work, we demonstrate our approach for a day-ahead planning problem. We extend our earlier work on probabilistic N-1 security, to incorporate recent results on convex AC optimal power flow relaxations. The problem is formulated as a chance constrained convex program; to deal with the chance constraint we follow an algorithm based on a combination of randomized and robust optimization. We also enhance the controllability of the system by introducing a corrective scheme that imposes post-contingency control of the Automatic Voltage Regulation (AVR) set-point. This scheme allows us to inherit a priori probabilistic guarantees regarding the satisfaction of the system constraints, unlike the base case where the AVR set-points are constant. To illustrate the performance of the proposed security-constrained AC optimal power flow we compare it against a DC power flow based formulation using Monte Carlo simulations, and show that it results to lower operational cost compared to the case where the AVR set-points are constant.