Optimal power flow with limited and discrete controls

Operators solve optimal power flow problems to determine how to adjust generator dispatches and control devices in order to maintain a power system running in a secure and efficient manner. However, optimal power flow software typically make use of a number of control adjustments that is impractical for an operator to execute due to time constraints. Moreover, the common rounding techniques for handling the discrete nature of control devices, in particular switched shunts, may result in unnecessarily poor solutions. These deficiencies are addressed by exploring the use of a sparsity-inducing penalty to obtain a more manageable number of control adjustments, and the use of a distributed line-search for exploring the space of discrete variables. The benefits and computational requirements of these techniques have been evaluated on two real North American power networks of approximately 2.5k buses.