Generalized Sigma approach to unit commitment with uncertain wind power generation

Abstract Using stochastic unit commitment (SUC) to study the integration of uncertain wind power in transmission congested systems is computationally very intensive. This paper therefore presents an efficient deterministic alternative to SUC that generalizes the traditional three-sigma approach to transmission congested systems. In this Generalized Sigma or G-Sigma method, the bus residual demands and line power flows are treated as correlated random variables. Compared with SUC where wind power forecast uncertainty is represented through a large pool of scenarios, each with a distinct probability of occurrence, in the proposed G-Sigma method, a few representative scenarios are chosen on the perimeter of the elliptical confidence region corresponding to a certain level of security. The generation set points and reserve capacities are then scheduled to meet these few representative scenarios in addition to the expected error-free scenarios, thus guaranteeing a minimum level of security. The accuracy and performance of the proposed G-Sigma method were gauged and compared to a benchmark SUC via a three-region multi-unit system and by the IEEE 24-bus reliability test system with multiple units.