 |
Center for Nonlinear Studies |
The increasing level of uncertainty in power systems highlights the need to revisit operational concepts and develop novel design methodologies to achieve a better trade-off between secure and economic operation. We first discuss the concept of probabilistic security and consider the optimal power flow problem with N-1 security constraints and renewable in-feed uncertainty. We consider both the DC and AC power flow formulations. We introduce a feedback policy on frequency and voltage control loops and show how these problems can be transformed into chance constrained optimization programs. To solve these programs, we demonstrate how recently developed algorithms based on uncertainty sampling that offer a-priori guarantees regarding the probability of constraint satisfaction can be exploited. These methods avoid assumptions on the distribution of the uncertainty. We then show how our stochastic security constrained optimal power flow framework can be extended to incorporate optimal reserve decisions and propose a reserve strategy that can be deployed in real time for any realization of the uncertainty. Simulation results are shown in IEEE benchmark systems and the efficiency on large-scale systems is discussed. We have also extended this framework in different directions e.g. including scheduling of uncertain reserves from demand response, unit commitment decisions, scheduling of other controllable network components, and controlling risk based on outage probabilities.
Host: Misha Chertkov